Ординатура / Офтальмология / Английские материалы / Adequate HLA Matching in Keratoplasty_Sundmacher_2003

Medawar [2] demonstrated the homology between tissue and leukocyte antigens. He showed that the rejection of skin transplants in rabbits was significantly accelerated when preceded by a previous skin transplant or by the injection of leukocytes.

It was not until 1962 that the first di-allelic human leukocyte group system, 4a and 4b, was described [3]. In subsequent years, knowledge of human leukocyte antigens has progressed to the state where many antigenic determinants controlled by multiple closely linked complex loci can be recognized by serologic and/or molecular techniques.

Similar studies in other vertebrate species have demonstrated the presence of clusters of genes on a single chromosomal segment that control similar antigens and other immunologically interrelated functions. Those genes constitute the major histocompatibility complex (MHC) of the species.

Frelinger and Shreffler [4] have listed the following common features of MHCs: (1) principal transplantation barrier of the species; (2) serologically detected antigens of lymphocytes, broadly distributed on other tissues; (3) major factors which stimulate mixed leukocyte reaction (MLR) and graft-versus-host reaction (GVHR); (4) immune response genes; resistance to disease; (5) multiple phenotypic traits or functions controlled by a tight cluster of multiple genetic loci, and (6) extensive genetic polymorphism at many loci in the complex.

The driving force behind the identification of the human major histocompatibility locus (HLA) was the search for polymorphic antigens to be used to match donors with recipients for transplantation.

There is nothing in the definition of MHCs that justifies the exclusion of corneal tissue from its effects. Nevertheless, that erroneous assumption has been accepted by some investigators since they were not able to demonstrate a beneficial HLA matching effect.

HLA typing and matching have been used for more than four decades for solid organ transplantations. However, that practice has not yet achieved universal acceptance for cornea transplantation, mainly because of inconclusive or contradictory analytical results [5–29].

Van Rood et al. [30] cited four possible reasons for disparate results in kidney transplantation follow-up studies: (1) poor quality HLA typing negates the effect of matching; (2) when the number of good (or poor) matches is low, a matching effect is not demonstrable; (3) aggressive immunotherapy might diminish the influence of HLA matching on graft survival (however, at the cost of other complications such as malignancies [31, 32]), and (4) racial heterogeneity can make it impossible to obtain good matches.

We avoided the adverse effects of those factors by performing a comprehensive study of the immune response to HLA-A, -B and -DR histoincompatibilities in 1,681 consecutive recipients of penetrating keratoplasties which were

performed by a single surgeon, in a single center, over a 20-year period, 1976–1996 [33].

Materials and Methods

Patients, HLA typings and their quality control, HLA-A, -B and -DR matching of recipients and donors, population genetics validation of HLA typing results, storage and quality control of donor corneas, clinical diagnostic criteria of graft rejection, post-operative follow-up and therapy, and data handling and statistical methods have been described previously. We also performed simulation studies to demonstrate the effects of mistyped recipient and donor HLA-DR typings on analytical results [33].

Irreversible immunological rejections constituted 41% of the causes of failure in our results. The non-immunological causes included glaucoma, recurrent herpes simplex virus infection, insufficient donor material, bacterial infections, secondary endothelial dystrophy and trauma. Cases with such failures were censored, i.e. excluded, from the analyses.

Survival curves were calculated using the actuarial life table method [34]. The significance of differences between classes was assessed with a 2 statistic derived from a log-rank test [35]. Relative risk (RR) estimates were derived from the values in the logrank tables.

The absence of any rejection episode, reversible or irreversible, is a more sensitive and appropriate indicator of the immune competence of corneal transplant recipients than the time to irreversible immunological rejection that is really an indication of the efficacy of follow-up procedures, especially the immunosuppressive therapy protocols, of transplantation centers. However, since that indicator is not used universally, we have performed all of our major analyses using both indicators to enable investigators to compare their results with ours, irrespective of the end-point indicator that they use.

A Cox proportional hazards model was used for the multivariate analyses. Failures due to non-immunological causes were censored in the statistical analyses [36].

For a recent review of the essentials of HLA typing and the procedures for validating its results, see Schreuder [37] and Schipper and D’Amaro [38].

Results

Precision of HLA-DR Typings

In 1986, the HLA typing laboratory which performed all of our recipient and donor typings compared its results for 964 donor HLA-DR typings, which had been obtained with a serological cytotoxicity test using an HLA-DR serum set provided by Eurotransplant and/or local sera, with the results which it had obtained by using a PCR-biotin-SSO technique which was performed on the mononuclear cells from donor spleen samples. Those results revealed a concordance rate for the serological and molecular HLA-DR typings of

Völker-Dieben/Schreuder/Claas/Doxiadis/Schipper/Pels/Persijn/Smits/D’Amaro 24

90.8% [39]. In 1997, the concordance rate for 382 HLA-DR retypings was 99.2% [40].

We used population genetic analytical techniques to confirm the results of the periodic quality assurance studies that are carried out regularly to assess the precision of the HLA typings. The goodness-of-fit tests for Hardy-Weinberg equilibrium for the HLA-A, -B and -DR loci were used to assess if the numbers of different phenotype combinations in those loci agreed with the numbers predicted by the gene frequencies of the HLA alleles in those loci [41, 42]. The test result p values were all clearly above the minimum significance level of 0.05. For the recipient typings, the p values were 0.385 for HLA-A, 0.111 for HLA-B and 0.906 for HLA-DR. For the donor typings, the p values were 0.747 for HLA-A, 0.116 for HLA-B and 0.833 for HLA-DR.

The HLA gene frequencies in the donor and recipient populations were similar to each other and to those in a healthy Dutch Caucasoid control population. Those precise typing results were not surprising since the same experienced laboratory typed all of those individuals, but they nevertheless demonstrate the consistency and reliability of the typing results that formed the basis for assigning the match grades that were used to pair donors with recipients.

Identification of Factors That Exert a Beneficial Effect on the Survival of Corneal Transplants

Monovariate Analyses

Monovariate analyses of high-risk cases, i.e. those with recipient corneal vascularization in 2 or more quadrants, were used to identify the factors that exerted a significant beneficial effect on the survival of corneal transplants. The actuarial survival curve analyses were performed twice. Their results are shown only for the significant factors. The following additional factors were tried also but their results were not significant: blood transfusion history; recipient and donor ABO blood groups, and age.

HLA-A, -B Matching: Only Irreversible Immunological Rejection

Episodes

The first set of survival curve analyses used only the high-risk cases whose cause of failure was only an irreversible immunological rejection event (table 1). The Kaplan-Meier estimates of the cumulative proportion of corneal grafts

Table 1. Results of monovariate analyses; failure is irreversible immunological rejection only

mism Mismatches; p value significance of the difference between the 2 survival curves.

aCases: For transplant number, 388/463 indicates 388 cases who were retransplanted and 463 cases who received only 1 transplant. A similar interpretation should be used for the cases listed for the other factors.

bNo selection for vascularization risk status. All cases analyzed.

which failed due to only irreversible immunological events are displayed in figure 1 for the 642 HLA-AB matched and 207 AB mismatched cases: 85.3% at 5 years for the 0–2 AB mismatched cases and 67.4% at 5 years for the 3–4 AB mismatched cases, p value by log-rank test 0.00003. The RR of failure for the 3–4 AB mismatched cases versus the 0–2 AB mismatched cases was 2.73, p 0.0001.

HLA-DR Matching: Only Irreversible Immunological

Rejection Episodes

The Kaplan-Meier estimates of the cumulative proportion of corneal grafts which failed due to only irreversible immunological rejection events are displayed in figure 2 for the 66 HLA-DR matched and 214 DR mismatched cases: 97.1% at 5 years for the 0 DR mismatched cases and 82.7% at 5 years for the 1–2 DR mismatched cases, p value by log-rank test 0.005. The RR of failure for the 1–2 DR mismatched cases versus the 0 DR mismatched cases was 8.06, p 0.005 (table 1).

Völker-Dieben/Schreuder/Claas/Doxiadis/Schipper/Pels/Persijn/Smits/D’Amaro 26

Fig. 1. Influence of HLA-AB matching on high-risk corneal transplants. Only irreversible immunological rejections were analyzed. High-risk recipients are those with moderate or severe pre-operative corneal vascularization in 2 or more quadrants.

Cases at risk at 6 month intervals after cornea transplantation:

HLA-A, -B Matching: All Immunological Rejection Episodes

The second set of survival curve analyses used our entire data set of high-risk recipients and their donors. All immunological events, reversible or irreversible, were analyzed (table 2). The Kaplan-Meier estimates of the cumulative proportion of corneal grafts which failed due to any immunological event are displayed in figure 3 for the 642 HLA-AB matched and 207 AB mismatched cases: 75.2% at 5 years for the 0–2 AB mismatched cases and 57.7% at 5 years for the 3–4 AB mismatched cases, p value by log-rank test 0.0004. The RR of failure for the 3–4 AB mismatched cases versus the 0–2 AB mismatched cases was 2.09, p 0.0001.

Fig. 2. Influence of HLA-DR matching on high-risk corneal transplants. Only irreversible immunological rejections were analyzed. High-risk recipients are those with moderate or severe pre-operative corneal vascularization in 2 or more quadrants.

Cases at risk at 6 month intervals after cornea transplantation:

HLA-DR Matching: All Immunological Rejection Episodes

The Kaplan-Meier estimates of the cumulative proportion of corneal grafts which failed due to any immunological event are displayed in figure 4 for the 66 HLA-DR matched and 214 DR mismatched cases: 84.8% at 5 years for the 0 DR mismatched cases and 71.4% at 5 years for the 1–2 DR mismatched cases, p value by log-rank test 0.040. The RR of failure for the 1–2 DR mismatched cases versus the 0 DR mismatched cases was 2.39, p 0.024 (table 2).

Völker-Dieben/Schreuder/Claas/Doxiadis/Schipper/Pels/Persijn/Smits/D’Amaro 28

Table 2. Results of monovariate analyses; failure is any immunological rejection episode, reversible or irreversible

mism Mismatches; p value significance of the difference between the 2 curves. aCases: For vascularization, 851/830 indicates 851 high-risk cases and 830 low-risk cases.

A similar interpretation should be used for the cases listed for the other factors. bNo selection for vascularization risk status. All cases analyzed.

Joint Effect of HLA-A, -B and -DR Matching

The hypothesis for a joint effect of HLA-A, -B and -DR matching was supported by the fact that, among the 553 cases that were typed for those three loci, 7 grafts with 0 DR mismatches but 1 or 2 AB mismatches failed and 8 grafts with 0 AB mismatches but 1 DR mismatch failed. A similar observation was made in 1986 by Boisjoly et al. [43].

Multivariate Analyses on the Effect of HLA Matching on the Survival of Corneal Transplants

The hypothesis of an effect of MHC class I, HLA-A, -B and MHC class II, HLA-DR mismatches on corneal graft survival was tested in a stratified Cox model using a backwards selection procedure based on the likelihood ratio test. Four analyses were performed.

The first analysis used only the time to irreversible immunological rejection as the entry time. The final Cox proportional hazards model identified HLA-DR matching, % panel-reactive antibodies, retransplantation, HLA-AB matching, number of immunological rejection events, (reversible or irreversible), and vascularization risk class. Its results and their significance (p 0.000001) are set out in table 3 (p. 32).

The second analysis used the time to any immunological rejection event, reversible or irreversible, as the entry time. We modeled the following clinically

Fig. 3. Influence of HLA-AB matching on high-risk corneal transplants. All immunological rejections, reversible or irreversible, were analyzed. High-risk recipients are those with moderate or severe pre-operative corneal vascularization in 2 or more quadrants.

Cases at risk at 6 month intervals after cornea transplantation:

relevant factors: time, degree of vascularization, retransplantation, number of HLA-AB and -DR mismatches, proportion of panel-reactive antibodies and number of immunological rejection events. Recipient and donor gender, endothelial cell count, storage medium, graft size, recipient and donor ABO blood groups and age were also tried but they did not make any significant contribution to the models. The results of the final Cox proportional hazards model that identified vascularization risk class, and HLA-AB and HLA-DR match grades, and their significance (p 0.000001) are set out in table 4 (p. 32).

The third and fourth analyses were similar to the first and second with the addition of stratification for the four consecutive 5-year intervals covered by

Völker-Dieben/Schreuder/Claas/Doxiadis/Schipper/Pels/Persijn/Smits/D’Amaro 30

Fig. 4. Influence of HLA-DR matching on high-risk corneal transplants. All immunological rejections, reversible or irreversible, were analyzed. High-risk recipients are those with moderate or severe pre-operative corneal vascularization in 2 or more quadrants.

Cases at risk at 6 month intervals after cornea transplantation:

our 20-year study. Their results were not appreciably different to those in the first two analyses. They demonstrated that the performance of our Center was consistent over those two decades (data not shown).

Simulation Studies to Assess the Effect of Imprecise HLA-DR Typings

The monovariate analyses of our corneal transplant results revealed a significant beneficial effect of HLA-A, -B and -DR matching (tables 1–4, fig. 1–4). Our results were markedly different from those in the report of the CCTS study [15]. The high degree of reproducibility of our donor and recipient HLA-DR typings made us realize that we could use our well-typed HLA-DR typing results

Table 3. Results from the Cox proportional hazards model; censoring indicator variable is irreversible immunological rejection (455 observations)

Likelihood ratio test: 40.4041, DF 6, p value 0.000001.

Table 4. Results from the Cox proportional hazards model; censoring indicator variable is any immunological rejection event – reversible or irreversible (544 observations)

Likelihood ratio test: 36.6135, DF 3, p value 0.000001.

to demonstrate the effect of typing errors on analytical results which are obtained from imprecise typing data. Our goal was to determine if the results of the CCTS study could have been caused by their reported serious level of HLA-DR typing discrepancies, which was almost 40% in their recipients who had only a single reported HLA-DR antigen [44]. To achieve that goal, we randomly introduced HLA-DR typing errors of 5–40% into the HLA-DR typing results of our 280 high-risk recipients and their donors in our data file. We examined the influence of those errors on the demonstrated significant beneficial effect of HLA-DR matching (tables 1, 2, fig. 2, 4). This simulation approach is not dependent on any particular typing technique since it simply uses the phenotype assignments that were based on typing results.

When the censoring indicator variable was only irreversible immunological rejection events, the difference between the proportion of surviving grafts in the HLA-DR mismatch classes (0 and 1–2 DR mismatches), 5 years after

Völker-Dieben/Schreuder/Claas/Doxiadis/Schipper/Pels/Persijn/Smits/D’Amaro 32